HP virtual connect Ethernet Module with HSRP

[Lawrence Chia Chao Xian]

Hi All,

I am facing a issue in using virtual connect (VC) with HSRP.

The setup is as below.

Router1              Router2

    |                         |

    | (Link 1)             | (Link2

    |                         |

VC1     -----------     VC2

    |                        |

    | ----- Blade1 ------|

             NIC1

Router1 is running HSRP with Router2, Link1 is connected from Router1 to VC1 and Link2 is connected from Router2 to VC2. Both the VCs are in the same enclosure and are connected via a internal chasis backplane. Currently, the link on the VC1 is showing active and the link on VC2 is showing standby. This is causing a problem as Router2 is not able to receive hello packets from Router1 which results in the HSRP not working properly.

Does anyone have similar setup, and is there any workaround to let Link2 become active so that Router2 can receive the hello packet from Router1 which will allow the HSRP to work properly?

Happy New Year in advance!

Regards,

[Reza Sharifi]

Hello Larry,

Is there a link between router 1 and router 2?

What type of enclose, IBM chassis, HP Chassis or stand alone server?

Also, the default gateways for the server is the routers with no switch in between right?

Reza

[Lawrence Chia Chao Xian]

Hi Reza,

There is a link between router1 and router2, its is a HP enclosure. The default gateway is to the router HSRP VIP with no Physical switch in between, we are using Virtual connects to act as a switch between the blades and the router.

Regards,

Lawrence

[Reza Sharifi]

Hi Larry,

Please take a look at this document.  There is no connectivity between VC-0 and VC-1, so I don't think the HSRP hello can go between theses two VCs unless you have another switch in between the routers and the VCs.

http://www.vmware.com/pdf/esx3_vlan_wp.pdf

HTH

Reza

[gnijs]

All,

Virtual Connect is not a switch. It will NEVER pass packets from one uplink to the other.

Virtual Connect does not run spanning tree, and therefore it is not allowed to pass packets from one uplink to the other as this would require STP for loop prevention.

If you have dual L2 switches upstream, you have to interconnect them directly between each other. HSRP will run on this interconnect link.

You will have to trust HP NOT to pass any packets between the active and the standby uplink

TIP: run STP on the links towards the HP chassis and put BPDUguard on it. If there is ever a bridging loop within the HP chassis (ie. a server guy configures MS bridging between the two server NICs !), BPDUguard will block both uplinks and isolate the HP chassis. At least the rest of your datacenter will continue to run

regards,

Geert

[seanmcge]

Not a switch? Will never allow packets between uplinks? Are you sure?

Virtual Connect Bug Fixes:

Loop #1: (Aug 2008) “Resolved a problem that caused a momentary network loop ...”
Loop #2: (Oct 2007) “Fixed a problem that allowed a brief network loop to be formed…”
(http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareDescription.jsp?lang=en&cc=us&prodTypeId=3709945&prodSeriesId=3552695&swItem=MTX-3118ef34eb9c4503924a52b34c&prodNameId=3552696&swEnvOID=1005&swLang=8&taskId=135&mode=5)

Regards,

-sean

[rvenzor2000]

You may need to set up two seperate uplink trunks, and assign connections from each VC to each of these. This would put both Virtual Connects to an Active/Active configuration.

Scenario 1.2 in the VC Cookbook talks about this.

http://h20000.www2.hp.com/bc/docs/support/SupportManual/c01990371/c01990371.pdf

Configure the uplinks the way the document tells you in the back of the book.

I hope this helps.